bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Zeb2 ChIP-seq in stem cells
Zeb2 DNA-binding sites in ES cell derived neuroprogenitor cells reveal autoregulation and align
with neurodevelopmental knockout mouse and disease phenotypes.
Judith C. Birkhoff1, Anne L. Korporaal1, Rutger W.W. Brouwer2, Claudia Milazzo1#, Lidia
Mouratidou1, Mirjam C.G.N. van den Hout2, Wilfred F.J. van IJcken1,2, Danny Huylebroeck1,3*, $,
Andrea Conidi1*, $
1 Dept. Cell Biology, Erasmus University Medical Center, Rotterdam, 3015 CN, The Netherlands;
2 Center for Biomics-Genomics, Erasmus University Medical Center, Rotterdam 3015 CN, The
Netherlands; 3 Dept. Development and Regeneration, KU Leuven, Leuven 3000, Belgium
* shared senior authors
# Present address: Dept. Clinical Genetics, Erasmus University Medical Center, Rotterdam, 3015 CN,
The Netherlands
$ Corresponding author:
Andrea Conidi, PhD, Dept. Cell Biology, Erasmus MC, Building Ee - room Ee-1040, Wytemaweg 80,
3015 CN Rotterdam, The Netherlands; Tel: +31-10-7043169; Fax: +31-10-7044743; E-mail:
1 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Author contributions
JB: Conception and design, collection and/or assembly of data, data analysis and interpretation,
manuscript writing
AK: Collection and/or assembly of data
RB: Data analysis and interpretation
CM: Collection and/or assembly of data
LM: Collection and/or assembly of data
MvdH: Collection and/or assembly of data, data analysis and interpretation
WvIJ: Data analysis and interpretation, manuscript writing
DH: Conception and design, financial support, manuscript writing, final approval of manuscript
AC: Conception and design, collection and/or assembly of data, data analysis and interpretation,
manuscript writing, final approval of manuscript
Key words
Embryonic stem cells; genome-wide binding sites; Mowat-Wilson syndrome; neuroprogenitor cells;
Zeb2
2 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Abstract
Perturbation and mechanistic studies have shown that the DNA-binding transcription factor Zeb2
controls cell fate decision, differentiation and/or maturation in multiple cell lineages in embryos and
after birth. In cultured embryonic stem cells (ESCs) Zeb2’s strong upregulation is necessary for their
exit from primed pluripotency and entering neural and general differentiation. We engineered mouse
ESCs to produce epitope-tagged Zeb2 from one of its two endogenous alleles. Using crosslinking
ChIP-sequencing, we mapped for the first time 2,432 DNA-binding sites of Zeb2 in ESC-derived
neuroprogenitor cells (NPCs). A new, major site maps promoter-proximal to Zeb2 itself, and its
homozygous removal demonstrates that Zeb2 autoregulation is necessary to elicit proper Zeb2 effects
in ESC→NPC differentiation. We then cross-referenced all Zeb2 DNA-binding sites with
transcriptome data from Zeb2 perturbations in ESCs, ventral forebrain in mouse embryos, and adult
neurogenesis. While the characteristics of these neurodevelopmental systems differ, we find
interesting overlaps. Collectively, these new results obtained in ESC-derived NPCs significantly add
to Zeb2’s role as neurodevelopmental regulator as well as the causal gene in Mowat-Wilson
Syndrome. Also, Zeb2 was found to map to loci mutated in other human congenital syndromes,
making variant or disturbed levels of ZEB2 a candidate modifier principle in these.
Significance statement
Zeb2 is needed for exit from primed pluripotency and differentiation of ESCs and being studied in
various cell lineages and tissues/organs in knockouts and adult mice. Phenotyping has mainly
documented transcriptomes of Zeb2-mutant affected cell types, but ChIP-seq data are still not
available to document Zeb2’s direct target loci as repressor or activator of transcription. This study
maps for the first time the Zeb2 genome-wide binding sites in ESC-derived neuroprogenitors, and
discovered Zeb2 autoregulation. Cross-referencing these ChIP-seq data with transcriptome profiles of
3 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Zeb2 perturbation further explains Mowat-Wilson Syndrome defects and proposes ZEB2 as candidate
modifier gene in other syndromes.
4 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Introduction
Zeb2 (also named Sip1/Zfhx1b) and Zeb1 (δEF1/Zfhx1a), the two members of the small family of Zeb
transcription factors (TFs) in vertebrates, bind to two separated E2-boxes CACCT(G), sometimes
CACANNT(G), on DNA via two highly conserved, separated clusters of zinc fingers [1-4]. Mutations
in ZEB2 cause Mowat-Wilson Syndrome (MOWS, OMIM#235730), a rare congenital disease with
intellectual disability, epilepsy/seizures and other defects [5-8]. Zeb2’s action mechanisms, partner
proteins, proven or candidate target genes, and genes whose normal expression depends on intact Zeb2
levels, have been studied in various cell types for explaining specific phenotypes caused by Zeb2
perturbation. Zeb2 DNA-binding around candidate direct target genes helped to explain Zeb2 loss-of-
function phenotypes in ESCs and cells of early and late embryos, and postnatal and adult mice. These
genes are involved in ESC pluripotency (Nanog, Sox2), cell differentiation (Id1, Smad7), and
maturation of various cell types, e.g., in embryonic cortical and adult neurogenesis (Ntf3, Sox6), as
well as epithelial-to-mesenchymal transition (EMT) (Cdh1) [9-16]. In reverse, subtle mutagenesis of
Zeb2 DNA-binding sites in demonstrated target genes has confirmed Zeb DNA-binding and its
repressive activity (on the mesoderm TF XBra and on epithelial Cdh1) [9, 17].
Despite its critical function in the precise spatial-temporal regulation of expression of
system/process-specific relevant genes during embryogenesis and postnatal development, but recently
also adult tissue homeostasis and stem cell based repair, and acute and chronic disease [18, 19], ChIP-
sequencing (ChIP-seq) data for Zeb2 has been obtained in very few cases only. This is the case for
high-Zeb2 hepatocellular carcinoma and leukemia cell lines, or cultured cells that overproduce tag-
Zeb2 from episomal vectors or the safe Rosa26 locus [15, 20-22], none of which represent normal,
endogenous Zeb2 levels nor express Zeb2 with normal dynamics. Furthermore, most if not all anti-
Zeb2 antibodies cross-react with Zeb1, so do not discriminate between both proteins when their
presence overlaps or succeeds to one another, and they compete for the same target genes, which for
the individual proteins depends on cell identity/state, extrinsic stimulation or cellular context (e.g., in
somitogenesis, in melanoma) [23,24].
In mouse (m) ESCs subjected to neural differentiation (ND), Zeb2 mRNA/protein is
undetectable in undifferentiated cells, whereas its strong upregulation accompanies efficient
5 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
conversion of naïve ESCs into epiblast stem cell like cells (EpiLSCs), and is essential for subsequent
exit from primed cells and progression to neuroprogenitor cells (NPCs) [15]. We have edited one Zeb2
allele of mESCs by inserting a Flag-V5-epitope coding sequence just before the stop codon, in-frame
with the last exon (ex9 of mouse Zeb2) [25]. These Zeb2-V5 mESCs were then differentiated to NPCs,
and Zeb2 binding sites determined by V5-tag Chip-Seq. We identified ~2,430 binding sites, of which
~1,950 map to protein-encoding genes. We then cross-referenced the ChIP-positive (+) protein-
encoding target genes with RNA-seq data of deregulated genes in cell-type specific
neurodevelopment-relevant Zeb2 perturbations [13, 25, 26]. Although we compare non-identical
systems, the ESC approach still revealed a number of interesting overlaps, as well as Zeb2’s role in
regulating critical targets in neurodevelopment. Taken together, we report for the first time the
identification of Zeb2 genome-wide binding sites in ESC neural differentiation.
Materials and Methods
Tag-Zeb2 mouse ESCs
gRNAs (Table 1) targeting Zeb2-ex9, and tracrRNA (Integrated DNA Technologies, IDT), were
diluted to 125 ng/µl in duplex buffer (IDT). gRNAs were annealed to tracrRNA at 1:1 ratio at 95°C for
5 min and cooling the samples to room temperature (RT, 24°C). 250 ng of these annealed gRNAs
were transfected in 350,000 mESCs together with 2 µg pX459-Cas9-puro vector and 1 µg ssDNA
oligo of the Donor Template containing the FlagV5-tag sequence (Table 1). Transfection was done in
a gelatin-coated 6-well plate using DNA:Lipofectamine2000 (ratio of 1:2). Six hours after transfection
the medium was refreshed, at 24 hours the cells were Puromycin-selected (2 µg/ml). After 2 days, the
remaining cells were transferred to gelatin-coated 10-cm dishes and given fresh ESC-medium (see
below). Per dish 1,000; 1,500; or 2,000 cells were plated and allowed to form colonies. Medium was
changed every other day. Colonies were picked, expanded and genotyped by PCR (both outer and
inner primer sets were used (Table S1; Fig. S1). All candidate clones were validated by Sanger-
sequencing; correct clones were expanded and validated by Western blot.
6 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
CRISPR/Cas9-mediated deletion of Zeb2 binding site located at chr2:45109746-45110421
Oligonucleotides for gRNAs (Table 1) with target outside of this chr2-region were cloned into BbsI-
digested pX330-hspCas9-T2A-eGFP plasmid. All plasmids were sequenced. 4 µg of gRNA-plasmids
(1 µg each) were transfected in 350,000 mESCs and selected (see above). After 24 hours these cells
were sorted as GFP+ cells (Becton-Dickinson LSR Fortessa). Per well of a 6-well plate 1,000; 1,500;
or 2,000 GFP+ cells were plated and colonies allowed to form, picked (see above) and genotyped by
PCR using primers flanking this deletion, and within and outside of it. Clones showing a possible
heterozygous or homozygous deletion, as concluded from the PCR analysis, were subjected to ND. At
D8, they were harvested, RNA was isolated and cDNA synthesized (see below), and amplified (for the
primers, see Table 1). All candidate clones were validated by Sanger-sequencing.
ChIP-seq
DNA libraries from input (i.e. control) and V5 ChIPs were prepared using ThruPLEX DNA protocol
(TakaraBio) specific for low amounts of DNA, and sequenced on Illumina HiSeq-2500, and single
reads of 50 bp were generated. Adapter sequences were trimmed from the 3’-end of the reads, after
which the reads were aligned to the mm10/GRCm38 genome using HISAT2 [74]. From the
alignments, secondary, supplementary, low quality and fragmented alignments (fragments > 150 bp)
were filtered away. Peaks were called with MACS [75], and coverage was determined. 42 and 25
million reads were generated for input and V5 ChIP, respectively.
ChIP-seq data analysis
Peak calling was performed with MACS2 (Galaxy version 2.1.1.20160309.6) [75, 76], with default
parameters (narrow peak calling, Mm1.87e9, FDR < 0.05) using the input sample as background. The
No model parameter was used, and the extension size was set on 210 bp based on the predicted
fragment lengths from the alignments ( MACS2 predict-tool, Galaxy version 2.1.1.20160309.1) [75,
76]. The distance of the aligned reads from the TSS of the gene was analyzed using ComputeMatrix
(Galaxy version 3.3.2.0.0) and PlotHeatmap Galaxy version 3.3.2.0.1; the used matrix is based on the
7 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
log2ratio of the aligned ChIP peaks over the input, calculated using BamCompare (Galaxy version
3.3.2.0.0) [77].
RNA-seq
The quality of total RNA (of biologically independent triplicates) of wild-type mESCs at D0, and at
ND-D4, D6 and D8, was checked on Agilent Technologies-2100 Bioanalyzer, using a RNA nano-
assay. All samples had RIN value of 9.8 or higher. Triplicate RNA-seq libraries were prepared
(Illumina TruSeq stranded mRNA protocol; www.illumina.com). Briefly, 200 ng of total RNA was
purified using polyT-oligo-attached magnetic beads for ending with polyA-RNA. The polyA-tailed
RNA was fragmented, and cDNA synthesized (SuperScript II, random primers, in the presence of
Actinomycin D). cDNA fragments were end-repaired, purified (AMPureXP beads), A-tailed using
Klenow exo-enzyme and dATP. Paired-end adapters with dual index (Illumina) were ligated to the A-
tailed cDNA fragments and purified (AMPureXP beads).
The resulting adapter-modified cDNAs were enriched by PCR (Phusion polymerase) as follows:
30 sec at 98°C, 15 cycles of (10 sec at 98°C, 30 sec at 60°C, 30 sec at 72°C), 5 min at 72°C. PCR
products were purified (AMPureXP beads) and eluted in 30 µl resuspension buffer. One μl was loaded
on an Agilent 2100 Bioanalyzer using a DNA-1000 assay to determine concentration and for quality
check. Cluster generation was performed according to the Illumina TruSeq SR Rapid Cluster kit v2
Reagents Preparation Guide (www.illumina.com). After hybridization of the sequencing primer,
sequencing-by-synthesis was performed using a HiSeq-2500 with a single-read 50-cycle protocol
followed by dual index sequencing. Illumina adapter sequences have been trimmed off the reads,
which were subsequently mapped against the GRCm38 mouse reference (using HiSat2 version 2.1.0)
[74]. Gene expression values were called (using HTSeq-count version 0.9.1) [78] and Ensembl release
84 gene and transcript annotation. Sample QC and DEG analysis have been performed in the R
environment for statistical computing (version 3.5.3, using DESeq2 version 1.22.1 and Tidyverse
version 1.2.1 (https://github.com/tidyverse; https://www.r-project.org/) [79, 80].
8 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Meta-analysis
RNA-seq datasets (as DEG tables; [13, 25]) were downloaded from GEO
(https://www.ncbi.nlm.nih.gov/geo/, GSE35616 and GSE103003, respectively). Cross-referencing and
visualization were performed in R, using Tidyverse, VennDiagram and pheatmap packages.
Pathways enrichment, GO, Function analysis and Gene-to-Disease association
These were performed with StringDB package for R, while for Gene-to-Disease association
Disgenet2R for R was used [81, 82].
Results
Heterozygous Zeb2-Flag-V5 ESCs differentiate as wild-type cells
Addition of short epitope(s) at the N- or C-terminus, as well as activation/repression domains of
heterologous TFs at the C-terminus, does not interfere with Zeb2’s DNA-binding (in Xenopus
embryos [27], heterologous cells [28], mouse forebrain [25], mESCs [15]). Here, we have successfully
used a CRISPR/Cas9 approach (Materials and Methods) to insert a Flag-V5-tag sequence in Zeb2-ex9
of mESCs (clone 2BE3; Fig. S1). Allele-specific RT-qPCR using primers that amplify sequences
between the ex9 and V5-tag showed mRNA expression from the tagged allele during ND at day (D) 6
and 8 (Fig. 1A). Western blot analysis confirmed in nuclear extracts of ND-ESCs at D8 (i.e., NPCs
[15]) the presence of Zeb2 of expected relative molecular mass, identified by anti-V5 and anti-Zeb2
antibodies (Fig. 1B). Both the Zeb2-V5 and wild-type ESCs were then verified for temporal
expression of Zeb2, core pluripotency genes (Pou5f1, Nanog, downregulated upon ND, and Sox2, the
latter also a neurogenesis TF) and an acknowledged NPC marker (Pax6) (Fig. 1C,D). Both cell lines
displayed similar gross expression dynamics of these genes, indicating that Zeb2-V5 ESCs at D8 of
ND can be used for ChIP-seq. Further confirmation in ND-ESCs came from selective pull-down of
Zeb2 on the known target Cdh1, using ChIP-qPCR. Zeb2 binds to 2 of 3 E-boxes in the Cdh1
promoter (Fig. 1E), which it represses during EMT [9. 15]. A 25-fold enrichment for Zeb2-V5 was
obtained when probing this same Cdh1 region, using anti-V5 antibody (αV5) conjugated beads (and
9 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
agarose beads as negative control; Fig. 1F). Hence, Zeb2-V5 binds to known Zeb2 target sites, and
these NPCs are suitable for mapping the Zeb2 genome-wide binding sites (GWBS) at normal,
endogenous levels of Zeb2.
One-third of the 2,432 Zeb2 DNA-binding sites in NPCs map close to the transcription start site
of transcriptome-confirmed, system-relevant protein-encoding genes, including Zeb2 itself
αV5-precipitated samples from upscaled Zeb2-V5 NPCs were used for cross-linking ChIP-seq,
followed by analysis with Galaxy Software [29]. Of the total of 2,432 significant peaks (for threshold
setting, Materials and Methods), 94% (2,294 peaks in total) mapped to loci (1,952 in total) that encode
protein, while 5% (125 peaks) map to micro-RNA genes (miRNAs), and 1% to regions lacking
annotation (ENSEMBL-GRCm38.99) (NA, Fig. 2A; File S1). About 37.5% of all binding sites are
located within -10/+10 kb of the annotated transcription start site (TSS, Fig. 2B). Gene ontology (GO)
pathway enrichment analysis of these 1,952 loci revealed binding of Zeb2 to classes of gene annotated
to Wnt, integrin, chemokine/cytokine (predominantly as defined in inflammation) and cadherin
signaling, respectively, and developmental signaling by EGF, VEGF, TGFβ and FGF family pathways
(Fig. 2C). Among these 1,952 genes, those for transcription regulatory proteins, post-translational
modification and metabolic enzymes, are well-represented (Fig. 2D).
In parallel we applied bulk temporal RNA-seq of wild-type mESCs at D0 (undifferentiated), D4
(neural induction), D6 (early NPCs) and D8 (NPCs) and checked the expression dynamics of the 1,952
Zeb2-bound genes (from the D8 sample); 1,244 of these were subject to significant transcriptional
regulation between D4-8 as compared to D0 (Fig. 2E; log2FoldChange <-0.5 or >0.5 and p.value <
0.05; low-stringency analysis was opted for to assess also small differences in mRNA of Zeb2-bound
genes). 335 of these genes are commonly expressed between D4-6-8, but at different levels (including
Zeb2 itself; for lists of all differentially expressed genes (DEGs), see File S2). Fig. S2A depicts the
D4, D6 and D8 transcriptomes of ND-mESCs, each compared to D0, with indication whether the
genes are bound or not by Zeb2, as determined by ChIP-seq. At each of the time points, hence at
different Zeb2 mRNA level, about 12% of the DEGs are bound by Zeb2 (Fig. S2B).
10 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Among the Zeb2-bound genes that are down regulated, Dnmt3l and Esrrb are present,
suggesting that upregulation of Zeb2 in ND-ESCs (D6 and D8) directly causes downregulation of
these two genes (Fig. S2C; File S2). Zeb2 has been suggested as direct repressor of Dnmt3l and Esrrb,
facilitating the switch from self-renewal of ESCs to their exit from pluripotency, and promotion of
differentiation. Indeed, levels of all Dnmt3 genes remained higher in Zeb2-knockout (KO) ND-ESCs
[15]. However, these Zeb2-KO cells convert less efficiently into EpiLSCs, and fail to exit from primed
pluripotency. Importantly, among the Zeb2-binding genes whose mRNA levels increased during ND,
Zeb2 itself is also present (yellow dot, Fig. S2C), indicating autoregulation. In fact, in this ND model
the highest recruitment of Zeb2-V5 in ChIP-seq data was mapped upstream the TSS of Zeb2 (File S1).
Out of the 1,244 Zeb2-bound genes that significantly changed steady-state mRNA levels in our
D4 to D8 transcriptome data sets, 213 are exclusive DEGs in NPCs at D8 (Fig. 2E; Fig. S2D). Among
these, Tcf4 is bound by Zeb2 and becomes highly upregulated in NPCs (Fig. 2E; Fig. S2D). Tcf4 is a
ubiquitous basic helix-loop-helix (bHLH) type TF that binds to E-boxes, and is active during central
nervous system (CNS) development [30, 31]. In oligodendrocyte precursors, Tcf4 is essential for their
subsequent differentiation; Tcf4 dimerizes with the lineage-specific bHLH-TF Olig2, further
promoting their differentiation and maturation [32], while Zeb2 together with upstream Olig1/2 are
essential for myelinogenesis in the embryonic CNS [11]. Here, Zeb2 generates anti-BMP(-Smad)/anti-
Wnt(-β-catenin) activities, which is crucial for embryonic CNS oligodendrocyte differentiation. The
regulatory action of ZEB2 on the TCF4 target gene suggested by our ChIP-seq may underpin
phenotypic similarities between MOWS and Pitt-Hopkins syndrome patients [33], the latter caused by
mutations in TCF4, making us speculate that TCF4 may be deregulated in MOWS neural cells.
Meta-analysis of RNA-seq data from selected neural-system Zeb2 perturbations and the NPC
ChIP-seq data reveal interesting overlaps
We performed a meta-analysis of three published transcriptome data sets from controls and Zeb2-KOs:
sorted E14.5 mouse ventral forebrain interneurons (Nkx2.1-Cre driven Zeb2-KO; [25]) and sorted
(day-2 after birth) progenitors of the V-SVZ, an adult neurogenic niche (Gsh2-Cre, see [25, 13]. In
addition, we used high-throughput RT-qPCR data generated on a Fluidigm platform and obtained after
11 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
esiRNA-based knockdown (KD) of Zeb2, as part of a systems-biology study in ND-mESCs (the Zeb2
KD data subset were provided by R. Dries [26]).
The DEGs upon these respective Zeb2 perturbations (p.value <0.05; log2FoldChange<-1 and
>1) were then filtered. This identified (i) genes that depend on normal Zeb2 levels for their
downregulation/repression (if directly by Zeb2, then as repressor) or (ii) other genes that depend on
Zeb2 for their upregulation/activation (if directly by Zeb2, as activator) (for Zeb2 as dual TF, see [19,
34]). In parallel, our 2,294 Zeb2-V5 sites in the 1,952 protein-encoding genes were filtered from the
complete ChIP-seq dataset, and then used as reference for the RNA datasets (Fig. 3A). This cross-
referencing identified 108 protein-encoding genes among the three transcriptomic and the ChIP-seq
datasets (Fig. 3A). Fig. S3 shows a heatmap of the changes in mRNA levels of these 108 genes in ND
of wild-type ESCs and their correlation with the analyzed datasets. Noteworthy, only Cxcr4 was
common in all RNA datasets. This is likely due to the two RNA-seq sets generated in different
brain/neuron cell-type in vivo mouse models, while the other RNA data documented the effects of
Zeb2-KD on mRNA levels of (only 96) TGFβ/BMP-system components [26]. Cxcr4 and its ligand
Cxcl12/Sdf-1 are crucial for migration of interneurons from the ventral forebrain to the neocortex [35,
36], processes co-controlled by Zeb2 [25]. These 108 genes are further involved in regulation of stem
cell pluripotency, signaling by TGFβ, FoxO and Hippo, and in axon guidance (Fig. 3B).
Zeb2 directly controls TGFβ/BMP-system component and neuronal differentiation/migration
genes
We then validated 14 out of the 108 cross-referenced target genes, either already known as Zeb2-target
(Nanog) or as TGFβ/BMP-system component (Bmp7, Tgfbr2, Smad1, Smad2, Smad3, Id2, Cited2) or
with crucial role in neurogenesis and neuronal maturation (Sema3f, Cxcr4, Lhx5, Ntng2, Pax6, Tcf4;
their mRNA levels in ND-ESCs are highlighted in the heatmap in Fig. S3). Because Zeb2-KO ESCs
do neither exit from primed pluripotency nor enter differentiation (Stryjewska et al., 2017), we chose
to validate our findings using shRNA-mediated Zeb2-KD at ND-D8 (shZeb2; Materials and Methods),
and analyzed these 14 genes two days later (D10) (Fig. 4A); at this read-out time point >50%
reduction of Zeb2 mRNA was obtained (Fig. 4B, top left panel).
12 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Zeb2-KD resulted in reduced mRNA levels of Cxcr4, Ntng2 and Pax6 (Fig. 4B, bottom row),
each involved in neuron specification and migration [37, 38]. Zeb2 -KD also caused down regulation
of Lhx5, involved in differentiation of interneurons, including cytoskeletal rearrangements during
dendritogenesis [39] and of Tcf4, involved in neurogenesis [30, 31]. Sema3f is an axon outgrowth and
neuronal guided migration cue, and was slightly upregulated (Fig. 4B). These results confirm the
regulation by Zeb2 of its direct targets in later phases of neuronal differentiation/migration; mutations
either in protein-coding or promoter/regulatory regions of these targets cause neurodevelopmental
disorders (see Discussion).
The expression of Nanog, the promoter of which binds Zeb2 repressor [15], was increased in
Zeb2-KD cells (Fig. 4B, upper row). Zeb2-KD caused increased mRNA of Bmp7, Tgfbr2, Smad1,
Smad2, Smad3 and Cited2 (Fig. 4B, upper/middle row), fitting with the normal levels of Zeb2
mounting anti-TGFβ/BMP family effects [19]. In contrast, Id2 is significantly downregulated in
shZeb2 ESCs (Fig. 4B, middle row). Id2 is normally activated by BMP-Smads and, together with
other Id proteins (Id1, Id3 and Id4), inhibits cell differentiation. e.g., Zeb2 represses Id2 in immune
cells to promote differentiation [12]. However, Id2 as well as other Id genes [40-42] is, like Zeb2 [10,
25], expressed in the developing forebrain. Taken together, these data suggest an active and direct role
for Zeb2 in repressing genes regulating stem cell pluripotency as well as a number of TGFβ/BMP-
system components (Bmp7, Smad1/2/3), but also in activating genes in neurogenesis (Cxcr4, Ntng2,
Lhx5).
Zeb2 potentiates its own gene expression, which is also crucial for proper control of its direct
target genes, but not for all
Strikingly, in our ChIP-seq dataset, the peak with the highest enrichment (~200-fold) mapped 232 bp
upstream of the Zeb2 TSS (Fig. 5A, File S1). We deleted this region (chr2:45109746-45110421) using
CRISPR/Cas9 in wild-type cells (Materials and Methods), thereby obtaining Zeb2ΔP/ΔP ESCs. Fig. 5B
(top left panel) shows that the Zeb2 mRNA levels in the homozygous ΔP clone stayed significantly too
low during ND, already from D8 onwards, compared to control cells. This classifies the Zeb2ΔP/ΔP
13 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
clone as a very successful alternative Zeb2-KD. We used Zeb2ΔP/ΔP mESCs to read-out the same genes
that depend on intact Zeb2 levels and are Zeb2 ChIP+ (see Fig. 4B). Levels of Zeb2 mRNA stayed
abnormally very low at D10 in Zeb2ΔP/ΔP ND-mESCs, wheereas Nanog was still expressed and
remained higher than in control cells (Fig. 5B, top row). Hence, Zeb2 levels, including those achieved
by autoregulation, are critical, but to a different degree, for proper regulation of neuronal-relevant
genes such as Cxcr4, Lhx5, Ntng2, Pax6, Tcf4 (Fig. 5B, bottom row). Among the TGFβ/BMP-system
components (see Fig. 4) we observed a slight reduction of Bmp7, Smad1 and Smad3, whereas Tgfbr2,
Δ Δ Smad2, Cited2 and Sema3f were not affected in Zeb2 P/ P ND-mESCs.
We speculate that Zeb2, and now including its novel autoregulation, plays an active role in
regulating a number of genes involved in neuron determination and, together with other TFs,
TGFβ/BMP-system component genes. Thus, the precise amounts of Zeb2 and in a critical stage also
its autoregulation are crucial in discriminating genes where Zeb2 plays the aforementioned primary,
active role (as for Cxcr4, Lhx5, Ntng2 etc). For these genes, ~50% reduction (as observed in the Zeb2-
KD) or the novel ΔP/ΔP “peak”mutation is sufficient to strongly deregulate them, but other genes are
either not or slightly affected (Sema3f, Smad2, Cited2 vs. Bmp7, Tgfbr2, Smad1, Smad3).
Because Zeb2 also binds phospho(p)-Smads [4, 11, 13, 19], we also scanned the Zeb2 ChIP+
direct target genes, and for which we saw strong deregulation upon Zeb2-KD and/or in Zeb2ΔP/ΔP cells
during ND (wherein Smad-activation is not stimulated), for the presence of (i) the Zeb half-sites
CACCT(G) (pragmatically neglecting the variable spacing between them) (ii) that combine with
T T A candidate p-Smad binding and responsive genes (using GTC( /G)CT( /G)( /C)GCC for p-
C Smad1/Smad5, GTCTAGAC for p-Smad2/3) and (iii) the co-Smad Smad4 (C( /T)AGAC) [43], using
the Jaspar database. Fig. S4 shows the distribution of such identified Zeb- and Smad-binding motifs
(threshold score >85%) in those genes strongly affected by Zeb2-KD and/or in Zeb2ΔP/ΔP cells.
Interestingly, in the regions where Zeb2 binds close to the TSS (Zeb2, Ntng2, Lhx5, Nanog), the p-
Smad and Smad4 binding elements are sometimes present in very close proximity of the ChIP+ Zeb2-
bound E-box.
14 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Discussion
We report for the first time the endogenous GWBS for Zeb2 by ChIP-seq, in ESC-derived NPCs
tailored for this. We have previously used Zeb2Δex7/Δex7-KO mESCs and, for rescue purposes, such cells
in which Flag3-Strep-Zeb2 was produced from a Cre-controllable Rosa26 locus [15, 44] .
Conceptually, with regard to Zeb2 levels, the latter cells are different from the mESCs established
now, for they had the drawback that Zeb2 is not subjected to its normal temporal regulation during
differentiation. However, precise dosage of Zeb2 is critical, as concluded from transgenic Zeb2
cDNA-based rescues in Zeb2-KO ESCs and similar genetic rescues in Zeb2-mutant cells in mice,
which can via heterozygous/homozygous combinations create a large panel of Zeb2 levels (in
interneurons [25], in NK cells [45], in ESCs [15]). Another illustration of fine-tuned control of Zeb2
levels are miRs (targeting Zeb2) and lncRNAs (that regulate these miRs) [46-49] , and Zeb2 on its turn
also controlling some of its own miR-encoding genes or clusters [50-52]. In our ChIP-seq we find 125
peaks (~5% of the total) that correspond to the TSSs of 98 miR-genes (File S1). Among these miR-
genes, Zeb2 binds to loci encoding miR-144, miR-148a, miR-9 and miR-153, known to target Zeb2 in
the context of e.g., tumor progression [53-56]. We recently added the identification, in human iPSCs
subjected to ND, of Zeb2 distant (~600 kb upstream) enhancers, which act through DNA-looping to
the ZEB2 promoter-proximal region [57]. In addition, we have documented dynamic expression
patterns of Zeb2 in early embryos [17, 23, 58-60]. We have also shown that cDNA-based expression
of various tag-Zeb2 proteins is compatible with functional embryology-type and action mechanism
studies [15, 27, 28]. Importantly, our Zeb2-V5 allele enables normal production of tag-Zeb2 for the
first time from its endogenous locus.
Only two studies present ZEB2 ChIP-seq data in human cells, i.e. SNU398 hepatocellular
carcinoma and K562 erythroleukemia cells, respectively [21, 22]. In K562 ZEB2 binds to the
promoters of NR4A2, NEUROG2 and PITX3, expressed in midbrain dopaminergic neurons, wherein
Zeb2 negatively regulates axon growth and target innervation [61]. In SNU398, ZEB2 represses the
GALNT3, which is normally expressed in epithelial cells. This repression co-incides with acquisition
of a mesenchymal phenotype, linking ZEB2 here again to an EMT-like process. These two valuable
studies also present limitations. The use of cancer cell lines of genomic instable nature may create
15 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
possible bias in ChIP-seq, and they overproduce ZEB2. Our ChIP-seq identifies >2,400 peaks for
Zeb2-V5 in mESCs at ND-D8. This is in cells not stimulated with TGFβ family ligands, with near-
38% of Zeb2 sites mapping close to the TSS (-10/+10kb). Most of these genes function in growth
factor/cytokine signaling and/or encode transcriptional regulators, the latter suggesting that Zeb2
orchestrates other (co)TFs driving the transcriptomic signature of NPCs. The regulation of Wnt
signaling by Zeb2 is in line with observations that inhibition of the Wnt-βcatenin pathway suppresses
ND in vitro and in vivo, and that Wnt (and Zeb2)-controlled Tcf4 expression promotes neurogenesis
and is required for normal brain development [62-66].
The 2,294 Zeb2 peaks map to 1,952 protein-coding genes, of which 1,244 are DEGs in ND-
mESCs. Strikingly, the strongest enrichment of Zeb2 occurs on the Zeb2 promoter itself, nicely
correlating with highest Zeb2 level during ND both in human iPSCs and mouse ESCs (data not
shown). These high levels depend on DNA-looping bringing three enhancers, located ~600 kb
upstream of the TSS, to the Zeb2 promoter [57], at least at early NPC state, but is then rapidly lost. We
have now identified a novel self-regulatory mechanism where Zeb2 binds upstream its TSS to
maintain its levels sufficiently high, also and at least during ND. While this autoregulation and/or
distant-enhancer mechanisms need further investigation in Zeb2-dependent differentiation and/or
maturation of other cell types (e.g. in cKO mouse models or MOWS patient cells), we propose that
lower Zeb2 levels might compromise this autoregulatory loop. When we deleted the Zeb2
autoregulatory site from both Zeb2 alleles (the ΔP/ΔP cells), Zeb2 mRNA levels drop significantly, but
Zeb2 is still expressed, and these cells can still exit from pluripotency and differentiate (data not
shown). A number of genes, which are mainly linked to neuron maturation, are significantly affected
Δ Δ in Zeb2 P/ P ESCs, whereas TGFβ/BMP-system component genes are not deregulated. Zeb2 dosage
might thus underlie this difference in regulating its direct, ChIP+ genes in our ESCs. Zeb2 might be
key to maintaining expression of neuronal genes, while for TGFβ/BMP-system genes Zeb2 may co-
operate with other TFs (including p-Smads) or DNA-modifying enzymes to regulate the expression of
targets.
Zeb2 binds to p-Smads, and several studies indicate its negative regulation of BMP-Smad
activation of specific target genes, although Zeb2 also has Smad-independent functions [13, 19].
16 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
BMP-pSmads bind to GGCGCC with high affinity [67]. Morikawa et al. [68] have confirmed these
results using ChIP-seq, but identified a lower-affinity (higher BMP-doses required) BMP-Smad
A element (GGAGCC). For achieving full responsiveness it was proposed that the GG( /C)GCC element
needs to be coupled with a Smad4 site, ideally located 5 bp away [68]. We find that in primary targets
affected by varying levels of Zeb2, E-boxes are located close to Smad-binding motifs. However,
whether Zeb2 and Smads are co-present in target regions requires further experiments, such as ChiP-
on-ChIP assays, and (non-neural) differentiation protocols (adding BMP and/or Nodal). However,
these may be further complicated because of post-translational modification status of Zeb2, nuclear p-
Smads and/or Smad4 [69-72].
Striking are also the 1,093 Zeb2-binding DEGs at D8. When we performed a gene-to-disease
association using the human orthologues of these D8-DEGs, we found a clear association with
disorders (Fig. S6). These include neurodevelopmental, mental and eye defects, which occur in
MOWS. Altogether, our data may provide novel insights in MOWS due to suboptimal ZEB2 amounts
in humans, and includes an autoregulation aspect from now on, and also that ZEB2 is a putative
modifier gene for many other congenital disorders.
Several Zeb2-cKO mouse models have been generated, and for many bulk RNA-seq data are
available. Here, we selected two such RNA-seq data sets [13, 25, 73]. In addition, similar data were
obtained for cultured mESCs, either Zeb2-KO cells [15] or cells submitted to ND wherein e.g., Zeb2-
KD was done [26]. Unfortunately, we cannot include Zeb2-KO mESCs, for they convert less
efficiently to EpiLSCs and fail to differentiate whatsoever [15]. The meta-analysis of these three
different data sets, overlaid with the 1,952 Zeb2-V5+ loci/genes, show therefore a limited number of
common targets, Cxcr4 being the only one. This narrows the Zeb2-bound gene collection to 108 in
total. However, and interestingly, the latter enrich for GO terms pluripotency of stem cells, signaling
by TGFβ and Wnt, cell fate commitment and neuron differentiation, all processes where Zeb2 plays a
crucial role. Out of these 108, we selected 14 genes covering TGFβ/BMP, pluripotency, and neuron
migration, differentiation/maturation, and checked their levels 2 days after Zeb2-KD at ND-D8. Most
of these 14 genes relevant to NPC status critically depend on intact levels of Zeb2. They may help to
explain why the defects caused by MOWS are observed later after birth and why (the few) missense
17 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
mutations in MOWS (besides the more abundant significant deletions etc) present with milder
syndromic manifestation. Both the cross-reference of Zeb2-ChIP+ genes with the transcriptome of
ND-ESCs and the meta-analysis identify a number of common genes, such as Bmp7, Tgfbr2, Tcf4,
Smad1/2/3, and Sema3f (File S3; Fig. S3), making Zeb2 a likely direct regulator of these. It is also
intriguing that Zeb2 is recruited to Tcf4, and controls it at D8, and that Tcf4 is deregulated upon Zeb2-
KD (esiRNA and shRNA). Mutations in TCF4 cause Pitt-Hopkins syndrome (PTHS, OMIM#610954),
a rare neurodevelopmental disorder with some defects overlapping with MOWS.
Conclusion and/or Summary
This ChIP-seq study in a suitable cell system identifies Zeb2 GWBS for the first time and discovers
Zeb2 autoregulation as a new principle, which may add to pathogenic mechanisms of MOWS, Zeb2-
executed control in TGFβ/BMP-Smad signaling of target genes, and its role as candidate modifier
gene in neurodevelopmental syndromes due to mutations in genes under direct ZEB2 control.
Acknowledgments
This work was supported mainly by Erasmus University Medical Center via departmental funds, extra
funds coordinated with its Executive Board, and BIG project funding, a collaborative extra support
between Erasmus University Rotterdam and its University Medical Center for promoting fundamental
research at the Theme of Biomedical Sciences, and hence the Department of Cell Biology. We thank
all co-workers of the Center for Biomics-Genomics at Erasmus University Medical Center for their
expert technical assistance, and colleagues Frank Grosveld and Raymond Poot for constructive
discussions.
Disclosure of Potential Conflicts of Interest
The authors declare no conflict of interest.
18 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Data Availability Statement
The data sets used and/or analyzed during the current study are available from the
corresponding author upon reasonable request.
19 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
References
1 Funahashi J, Kamachi Y, Goto K et al. Identification of nuclear factor deltaEF1 and its binding site essential for lens-specific activity of the delta 1-crystallin enhancer. Nucleic Acids Res 1991;19:3543-3547. 2 Sekido R, Murai K, Funahashi J et al. The delta-crystallin enhancer-binding protein deltaEF1 is a repressor of E2-box-mediated gene activation. Mol Cell Biol 1194;14:5692-5700. 3 Remacle JE, Kraft H, Lerchner W et al. New mode of DNA binding of multi-zinc finger transcription factors: deltaEF1 family members bind with two hands to two target sites. EMBO J 1999;18:5073-5084. 4 Verschueren K, Remacle JE, Collart C et al. SIP1, a novel zinc finger/homeodomain repressor, interacts with Smad proteins and binds to 5'-CACCT sequences in candidate target genes. J Biol Chem 1999;274:20489-20498. 5 Garavelli L, Donadio A, Zanacca C et al. Hirschsprung disease, mental retardation, characteristic facial features, and mutation in the gene ZFHX1B (SIP1): confirmation of the Mowat- Wilson syndrome. Am J Med Genet A 2003;116A:385-388. 6 Zweier C, Thiel CT, Dufke A et al. Clinical and mutational spectrum of Mowat-Wilson syndrome. Eur J Med Genet 2005;48:97-111. 7 Garavelli L, Ivanovski I, Caraffi SG et al. Neuroimaging findings in Mowat-Wilson syndrome: a study of 54 patients. Genet Med. 2017;19:691-700. 8 Ivanovski I, Djuric O, Caraffi SG et al. Phenotype and genotype of 87 patients with Mowat- Wilson syndrome and recommendations for care. Genet Med 2018;20:965-975. 9 Comijn J, Berx G, Vermassen P et al. The two-handed E box binding zinc finger protein SIP1 downregulates E-cadherin and induces invasion. Mol Cell 2001;7:1267-1278. 10 Seuntjens E, Nityanandam A, Miquelajauregui A et al. Sip1 regulates sequential fate decisions by feedback signaling from postmitotic neurons to progenitors. Nat Neurosci 2009; 12: 1373-1380. 11 Weng Q, Chen Y, Wang H et al. Dual-mode modulation of Smad signaling by Smad- interacting protein Sip1 is required for myelination in the central nervous system. Neuron 2012;73:713-728. 12 Scott CL, Soen B, Martens L et al. The transcription factor Zeb2 regulates development of conventional and plasmacytoid DCs by repressing Id2. J Exp Med 2016;213:897-911. 13 Deryckere A, Stappers E, Dries R et al. Multifaceted actions of Zeb2 in postnatal neurogenesis from the ventricular-subventricular zone to the olfactory bulb. Development 2020;147:dev184861. 14 Wu LM, Wang J, Conidi A et al. Zeb2 recruits HDAC-NuRD to inhibit Notch and controls Schwann cell differentiation and remyelination. Nat Neurosci 2016;19:1060-1072. 15 Stryjewska A, Dries R, Pieters T et al. Zeb2 Regulates Cell Fate at the Exit from Epiblast State in Mouse Embryonic Stem Cells. Stem Cells 2017;35:611-625. 16 Menuchin-Lasowski Y, Dagan B, Conidi A et al. Zeb2 regulates the balance between retinal interneurons and Muller glia by inhibition of BMP-Smad signaling. Dev Biol 2020;468: 80-92. 17 Lerchner W, Latinkic BV, Remacle JE et al. Region-specific activation of the Xenopus brachyury promoter involves active repression in ectoderm and endoderm: a study using transgenic frog embryos. Development 2000;127:2729-2739. 18 van Grunsven LA, Taelman V, Michiels C et al. deltaEF1 and SIP1 are differentially expressed and have overlapping activities during Xenopus embryogenesis. Dev Dyn 2006;235:1491- 1500.
20 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
19 Conidi A, Cazzola S, Beets K et al. Few Smad proteins and many Smad-interacting proteins yield multiple functions and action modes in TGFbeta/BMP signaling in vivo. Cytokine Growth Factor Rev 2011;22:287-300. 20 ENCODE Project Consortium. An integrated encyclopedia of DNA elements in the human genome. Nature 2012;489:57-74. 21 Balcik-Ercin P, Cetin M, Yalim-Camci I et al. Genome-wide analysis of endogenously expressed ZEB2 binding sites reveals inverse correlations between ZEB2 and GalNAc-transferase GALNT3 in human tumors. Cell Oncol (Dordr) 2018;41:379-393. 22 Yang S, Toledo EM, Rosmaninho P et al. A Zeb2-miR-200c loop controls midbrain dopaminergic neuron neurogenesis and migration. Commun Biol 2018;1:75. 23 Miyoshi T, Maruhashi M, Van De Putte T et al. Complementary expression pattern of Zfhx1 genes Sip1 and deltaEF1 in the mouse embryo and their genetic interaction revealed by compound mutants. Dev Dyn 2006;235:1941-1952. 24 Vandamme N, Denecker G, Bruneel K et al. The EMT Transcription factor ZEB2 promotes proliferation of primary and metastatic melanoma while suppressing an invasive, mesenchymal-like phenotype. Cancer Res 2020;80: 2983-2995. 25 Nelles L, Van de Putte T, van Grunsven L et al. Organization of the mouse Zfhx1b gene encoding the two-handed zinc finger repressor Smad-interacting protein-1. Genomics 2003;82:460- 469. 25 van den Berghe V, Stappers E, Vandesande B et al. Directed migration of cortical interneurons depends on the cell-autonomous action of Sip1. Neuron 2013;77:70-82. 26 Dries R, Stryjewska A, Coddens K et al. Integrative and perturbation-based analysis of the transcriptional dynamics of TGFbeta/BMP system components in transition from embryonic stem cells to neural progenitors. Stem Cells 2020;38:202-217. 27 Papin C, van Grunsven LA, Verschueren K et al. Dynamic regulation of Brachyury expression in the amphibian embryo by XSIP1. Mech Dev 2002;111:37-46. 28 Verstappen G, van Grunsven LA, Michiels C et al. Atypical Mowat-Wilson patient confirms the importance of the novel association between ZFHX1B/SIP1 and NuRD corepressor complex. Hum Mol Genet 2008;17:1175-1183. 29 Afgan E, Baker D, Batut B et al. The Galaxy platform for accessible, reproducible and collaborative biomedical analyses: 2018 update. Nucleic Acids Res 2018;46:W537-W544. 30 Corneliussen B, Thornell A, Hallberg B et al. Helix-loop-helix transcriptional activators bind to a sequence in glucocorticoid response elements of retrovirus enhancers. J Virol 1991;65:6084-6093. 31 Forrest MP, Hill MJ, Quantock AJ et al. The emerging roles of TCF4 in disease and development. Trends Mol Med 2014;20:322-331. 32 Fu H, Cai J, Clevers H et al. A genome-wide screen for spatially restricted expression patterns identifies transcription factors that regulate glial development. J Neurosci 2009;29:11399-11408. 33 Brockschmidt A, Filippi A, Charbel Issa P et al. Neurologic and ocular phenotype in Pitt- Hopkins syndrome and a zebrafish model. Hum Genet 2011;130:645-655. 34 Hegarty SV, Sullivan AM, O'Keeffe GW. Zeb2: A multifunctional regulator of nervous system development. Prog Neurobiol 2015;132:81-95. 35 Stumm RK, Zhou C, Ara T et al. CXCR4 regulates interneuron migration in the developing neocortex. J Neurosci 2003;23:5123-5130. 36 Nash B, Meucci O. Functions of the chemokine receptor CXCR4 in the central nervous system and its regulation by mu-opioid receptors. Int Rev Neurobiol 2014;118:105-128.
21 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
37 Georgala PA, Carr CB, Price DJ. The role of Pax6 in forebrain development. Dev Neurobiol 2011;71:690-709. 38 Dias CM, Punetha J, Zheng C et al. Homozygous missense variants in NTNG2, encoding a presynaptic Netrin-G2 adhesion protein, lead to a distinct neurodevelopmental disorder. Am J Hum Genet 2019;105:1048-1056. 39 Lui NC, Tam WY, Gao C et al. Lhx1/5 control dendritogenesis and spine morphogenesis of Purkinje cells via regulation of Espin. Nat Commun 2017;8:15079. 40 Tzeng SF, de Vellis J. Id1, Id2, and Id3 gene expression in neural cells during development. Glia 1998;24: 372-381. 41 Blomfield IM, Rocamonde B, Masdeu MDM et al. Id4 promotes the elimination of the pro- activation factor Ascl1 to maintain quiescence of adult hippocampal stem cells. Elife 2019;8:e48561. 42 Havrda MC, Harris BT, Mantani A et al. Id2 is required for specification of dopaminergic neurons during adult olfactory neurogenesis. J Neurosci 2008;28:14074-14086. 43 Hill CS. Transcriptional Control by the SMADs. Cold Spring Harb Perspect Biol 2016;8:a022079. 44 Higashi Y, Maruhashi M, Nelles L et al. Generation of the floxed allele of the SIP1 (Smad- interacting protein 1) gene for Cre-mediated conditional knockout in the mouse. Genesis 2002;32:82- 84. 45 van Helden MJ, Goossens S, Daussy C et al. Terminal NK cell maturation is controlled by concerted actions of T-bet and Zeb2 and is essential for melanoma rejection. J Exp Med 2015;212:2015-2025. 46 Guan J, Liu P, Wang A et al. Long noncoding RNA ZEB2AS1 affects cell proliferation and apoptosis via the miR1225p/PLK1 axis in acute myeloid leukemia. Int J Mol Med 2020;46:1490-1500. 47 Jiang Y, Wu K, Cao W et al. Long noncoding RNA KTN1-AS1 promotes head and neck squamous cell carcinoma cell epithelial-mesenchymal transition by targeting miR-153-3p. Epigenomics 2020;12:487-505. 48 Yao H, Hou G, Wang QY et al. LncRNA SPRY4IT1 promotes progression of osteosarcoma by regulating ZEB1 and ZEB2 expression through sponging of miR101 activity. Int J Oncol 2020;56:85-100. 49 Cheng H, Zhao H, Xiao X et al. Long non-coding RNA MALAT1 upregulates ZEB2 expression to promote malignant progression of glioma by attenuating miR-124. Mol Neurobiol 2021;58:1006-1016. 50 Brabletz S, Brabletz T. The ZEB/miR-200 feedback loop--a motor of cellular plasticity in development and cancer? EMBO Rep 2010;11:670-677. 51 Exposito-Villen A, A EA, Franco D. Functional role of non-coding RNAs during epithelial-to- mesenchymal transition. Noncoding RNA 2018;4:14. 52 Gregory PA, Bracken CP, Bert AG et al. MicroRNAs as regulators of epithelial-mesenchymal transition. Cell Cycle 2008;7:3112-3118. 53 Guan H, Liang W, Xie Z et al. Down-regulation of miR-144 promotes thyroid cancer cell invasion by targeting ZEB1 and ZEB2. Endocrine 2015;48:566-574. 54 Pan Y, Zhang J, Fu H et al. miR-144 functions as a tumor suppressor in breast cancer through inhibiting ZEB1/2-mediated epithelial mesenchymal transition process. Onco Targets Ther 2016;9:6247-6255. 55 Nourmohammadi B, Tafsiri E, Rahimi A et al. Expression of miR-9 and miR-200c, ZEB1, ZEB2 and E-cadherin in non-small cell lung cancers in Iran. Asian Pac J Cancer Prev 2019;20:1633- 1639.
22 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
56 Wahab NA, Othman Z, Nasri NWM et al. Inhibition of miR-141 and miR-200a increase DLC-1 and ZEB2 expression, enhance migration and invasion in metastatic serous ovarian cancer. Int J Environ Res Public Health 2020;17:2766. 57 Birkhoff JC, Brouwer RWW, Kolovos P et al. Targeted chromatin conformation analysis identifies novel distal neural enhancers of ZEB2 in pluripotent stem cell differentiation. Hum Mol Genet 2020;29:2535-2550. 58 van Grunsven LA, Papin C, Avalosse B et al. XSIP1, a Xenopus zinc finger/homeodomain encoding gene highly expressed during early neural development. Mech Dev 2000;94:189-193. 59 Van de Putte T, Maruhashi M, Francis A et al. Mice lacking ZFHX1B, the gene that codes for Smad-interacting protein-1, reveal a role for multiple neural crest cell defects in the etiology of Hirschsprung disease-mental retardation syndrome. Am J Hum Genet 2003;72:465-470. 60 Takagi T, Nishizaki Y, Matsui F et al. De novo inbred heterozygous Zeb2/Sip1 mutant mice uniquely generated by germ-line conditional knockout exhibit craniofacial, callosal and behavioral defects associated with Mowat-Wilson syndrome. Hum Mol Genet 2015;24:6390-6402. 61 Hegarty SV, Wyatt SL, Howard L et al. Zeb2 is a negative regulator of midbrain dopaminergic axon growth and target innervation. Sci Rep 2017;7:8568. 62 Hirabayashi Y, Itoh Y, Tabata H et al. The Wnt/beta-catenin pathway directs neuronal differentiation of cortical neural precursor cells. Development 2004;131:2791-2801. 63 Slawny NA, O'Shea KS. Dynamic changes in Wnt signaling are required for neuronal differentiation of mouse embryonic stem cells. Mol Cell Neurosci 2011;48:205-216. 64 Li M, Santpere G, Imamura Kawasawa Y et al. Integrative functional genomic analysis of human brain development and neuropsychiatric risks. Science 2018;362:eaat7615. 65 Li H, Zhu Y, Morozov YM et al. Disruption of TCF4 regulatory networks leads to abnormal cortical development and mental disabilities. Mol Psychiatry 2019;24:1235-1246. 66 Mesman S, Bakker R, Smidt MP. Tcf4 is required for correct brain development during embryogenesis. Mol Cell Neurosci 2020;106:103502. 67 BabuRajendran N, Palasingam P, Narasimhan K et al. Structure of Smad1 MH1/DNA complex reveals distinctive rearrangements of BMP and TGF-beta effectors. Nucleic Acids Res 2010;38:3477-3488. 68 Morikawa M, Koinuma D, Tsutsumi S et al. ChIP-seq reveals cell type-specific binding patterns of BMP-specific Smads and a novel binding motif. Nucleic Acids Res 2011;39:8712-8727. 69 Liu S, Long J, Yuan B et al. SUMO Modification Reverses Inhibitory Effects of Smad Nuclear Interacting Protein-1 in TGF-beta Responses. J Biol Chem 2016;291:24418-24430. 70 Xu P, Lin X, Feng XH. Posttranslational Regulation of Smads. Cold Spring Harb Perspect Biol 2016;8. 71 Kim YJ, Kang MJ, Kim E et al. O-GlcNAc stabilizes SMAD4 by inhibiting GSK-3beta- mediated proteasomal degradation. Sci Rep 2020;10:19908. 72 Lin X, Wang Y, Jiang Y et al. Sumoylation enhances the activity of the TGF-beta/SMAD and HIF-1 signaling pathways in keloids. Life Sci 2020;255:117859. 73 McKinsey GL, Lindtner S, Trzcinski B et al. Dlx1&2-dependent expression of Zfhx1b (Sip1, Zeb2) regulates the fate switch between cortical and striatal interneurons. Neuron 2013;77:83-98. 74 Kim D, Langmead B, Salzberg SL. HISAT: a fast spliced aligner with low memory requirements. Nat Methods 2015;12:357-360. 75 Zhang Y, Liu T, Meyer CA et al. Model-based analysis of ChIP-Seq (MACS). Genome Biol 2008;9: R137.
23 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
76 Feng J, Liu T, Qin B et al. Identifying ChIP-seq enrichment using MACS. Nat Protoc 2012;7:1728-1740. 77 Ramirez F, Ryan DP, Gruning B et al. deepTools2: a next generation web server for deep- sequencing data analysis. Nucleic Acids Res 2016;44:W160-165. 78 Anders S, Pyl PT, Huber W. HTSeq--a Python framework to work with high-throughput sequencing data. Bioinformatics 2015;31:166-169. 79 R Core Team, 2018. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.r-project.org/ 80 Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA- seq data with DESeq2. Genome Biol 2014;15:550. 81 Szklarczyk D, Gable AL, Lyon D et al. STRING v11: protein-protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Res 2019;47:D607-D613. 82 Pinero J, Ramirez-Anguita JM, Sauch-Pitarch J et al. The DisGeNET knowledge platform for
disease genomics: 2019 update. Nucleic Acids Res 2020;48:D845-D855.
24 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Figure legends
Figure 1. Characterization of the heterozygous Zeb2-Flag-V5 mESC line and ChIP-qPCR validation. A) Allele-specific RT-qPCR using two sets of primers located either in exon7 (and therefore able to detect the whole Zeb2 mRNA produced by both alleles; orange bar) or located in exon9 and the V5- tag (thus recognizing specifically the knocked-in tagged allele; light blue bar). B) Western blot analysis showing V5 epitope containing Zeb2 in ES cell derived NPCs (at D8 of neural differentiation, ND) in nuclear extracts (NE), but not in cytoplasmic extracts (CE). Membranes were blotted with anti-V5 antibody (left panel, αV5) or anti-Zeb2 antibody (right panel, αZeb2 [10]. As control, a fraction of Zeb2-rich extract obtained from HeLa cells transfected with pcDNA3- V5Zeb2 vector was also separated in the same gel. C) Zeb2 mRNA levels in wild-type (WT, orange bar) and Zeb2-V5 (grey bar, clone 2BE3, indicated as Zeb2-V5) mESCs during ND. D) Pluripotency markers Nanog, Pou5f1 (Oct4) and Sox2 are down regulated in Zeb2-V5 mESCs similarly to WT. The neuronal marker Pax6 is also significantly upregulated during differentiation, like in WT mESCs. E) Scheme of the mouse Cdh1 promoter showing the three E-boxes located upstream the ATG start codon. Zeb2 binds specifically to only two of these, indicated as R1 [15]. F) ChIP-qPCR showing enrichment for Zeb2-V5 binding to the R1 region of the Cdh1 promoter. Agarose beads were used as negative control (in grey).
Figure 2. Zeb2V5 protein is mainly recruited at the TSS of transcriptional regulator encoding genes, predominantly those classified in Wnt signaling. A) 2,432 peaks were selected from our ChIP-seq data set (see Materials and Methods). Of these, 94% are associated with protein coding loci, 5% with miRNAs and the remaining 1% map to regions without functional annotation (NA). B) Frequency plot showing the binding of Zeb2V5 at and around (-10 to +10 kb) the TSS. C) The 2,294 peaks map to 1,952 protein encoding genes, many of which operate in Wnt signaling (File S1) or are (shown in D) transcriptional regulators. E) Of the 1,952 protein encoding genes, 1,244 are differentially expressed during ND, when compared to the undifferentiated state (at D0). Of these 1,244 genes, 335 are differentially expressed at all three time points of ND. A few examples are listed of DEGs uniquely expressed at one time point, as well as these that are shared among three time points; a full list is provided in File S2.
25 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Figure 3. Schematic representation of the meta-analysis Zeb2-bound genes versus RNA-seq datasets. A) 108 genes bound by Zeb2 are also differentially expressed in the three datasets from other studies in mouse models and ESCs (see text for details). Cxcr4 is the only DEG bound by Zeb2 and common among the three datasets. B) These 108 genes mainly map to signaling pathways regulating stem cell pluripotency, and effects of TGFβ family, FoxO, and Hippo signaling/activity.
Figure 4. shRNA-mediated KD of Zeb2 discriminate between primary and secondary target genes. A) Schematic overview of the shRNA transfection targeting Zeb2 (shZeb2) and read-out of the effect. Cellular aggregates at D8 of ND are dissociated and transfected with shZeb2 or against a scrambled, control sequence (shCTRL). Read-out is done two days after the start of shRNA addition. The list of shRNAs is given in Table S2. B) Zeb2 levels after KD were reduced to 40-50% of their normal level (shZeb2, orange bars) compared to shCTRL (blue bars). Bmp7, Cited2, Nanog, Sema3f, Smad1, Smad2, Smad3, and Tgfbr2 were significantly upregulated following Zeb2 KD, whereas genes encoding for neuronal specification and migration (Cxcr4, Lhx5, Ntng2, Pax6 and Tcf4) were downregulated.
Figure 5. Deletion of the Zeb2-binding, candidate autoregulatory site (chr2:45109746-45110421) impairs Zeb2 mRNA levels and neuronal markers. A) Schematic overview of the log(FoldEnrichment) of the peaks identified by ChIP-seq located in the mouse Zeb2 locus, and localization on top of Zeb2 intron/exon structure. The highest peak is located 232 bp upstream of the first translated exon. Grey arrow indicates the TSS of Zeb2. Δ Δ B) Zeb2 mRNA levels are strongly reduced in the Zeb2 P/ P clone. Expression levels (mRNA) of the target genes validated with shRNA (Figure 4). Most, but not all of the genes found to be affected Δ Δ following Zeb2 KD are also deregulated in Zeb2 P/ P mESCs in particular Id2 and the neuronal markers Cxcr4, Lhx5, Ntng2, Pax6 and Tcf4.
26 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Tables
Table 1: gRNAs and donor template used for CRISPR/Cas9-mediated Zeb2 editing.
Name Sequence CRISPR/Cas9 FlagV5 aaaatggaaaccaaatcagaccacgaagaagacaatatggaa Donor Template gatggcatcgaaGACTACAAAGACGATGACG ACAAGgatatcGGTAAGCCTATCCCTAAC In-frame knock-in of Flag-V5 tag CCTCTCCTCGGTCTCGATTCTACGTAAa ctactgcattttaagcttcctattttttttttccagtagtattgtt gRNA_ex9_1 GGAAACCAAATCAGACCACGAGG
gRNA_ΔZP1 CCCGCGCGCGTTTCAATGGGCGC gRNA_ΔZP2 CCCTCGCGAGTGCAACACACCAA Zeb2 peak deletion gRNA_ΔZP3 GGGCTCGGAGCGCTGCCGATCGG gRNA_ΔZP4 CCGCTGGACCGGGGGGGAGTTGA
Donor template: lowercase: homology arms located in exon 9 and 3’-UTR of Zeb2, underlined lowercase: mutated PAM sequences, uppercase: FLAG coding sequence, lowercase italics bold: EcoRI restriction site, underlined uppercase: V5 coding sequence, bold uppercase: STOP codon.
27 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Supplemental figures Figure S1. Characterization of Zeb2-V5 mESC line. A) Schematic overview of the design strategy, including showing also primers used for genotyping and detection of tagged allele. B) Genotyping results (selected part) showing the heterozygous band (red arrow) present in mouse (m) ESC clone 2BE3. Lower band represents the wild-type (WT) allele. C) Zeb2-V5 specific PCR showing the presence of the tagged allele only in clone 2BE3, and not in WT genomic DNA material. Clone 1BD4 was also used as negative control. D) Sanger DNA-sequencing results of clone 2BE3 showing the alignment with the Zeb2V5 sequence designed in silico. The last three amino acids (i.e. GME) of WT Zeb2 have been modified in DK to remove a PAM sequence and mutate the remaining PAM sequence, target of the gRNA, to avoid multiple cutting of the CAS9. An EcoRI restriction site was inserted between the Flag and the V5 coding sequences to facilitate for screening and an artificial STOP codon was added after the V5- coding sequence.
Figure S2. Cross-reference of Zeb2Flagv5-bound protein-coding genes and transcriptome of differentiating mESCs. A) Volcano plots showing the distribution of Zeb2 bound genes in the whole transcriptomes of D4, D6 and D8 neural differentiated mESCs. Red dots depict the genes bound by Zeb2 and grey crosses those not bound. B) About 12% of the up- or down-regulated genes are bound by Zeb2 (red bars) at D8 and the percentage does not change when expanding our analysis to the other points of differentiation. C) Of the 1,952 genes bound by Zeb2 and found being differentially expressed during mESC differentiation, 335 are in common among the three considered time points. Zeb2 itself is among these common genes and its expression increases during differentiation (yellow dot). The volcano plots show the distribution of the common genes during differentiation. D) Timepoint specific DEGs depicted as volcano plots. The grey arrow in D8 panel indicates Tcf4.
Figure S3. Expression of genes resulting from the meta-analysis of Zeb2-bound genes vs three independent RNA-seq datasets. Heatmap visualizes the log2FoldChange of the resulting genes during mESC differentiation. RNA-seq datasets where the genes have been found to be differentially expressed are annotated as green (Nkx2.1|Zeb2), purple (Zeb2 KD in mESCs) or yellow (Gsh2|Zeb2) squares.
28 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Figure S4. Distribution of Zeb binding sites (performed as single E-boxes; for details, see text) and TGFβ/BMP activated (phospho-)Smads and Smad4 binding elements within the identified Zeb2- bound regions for primary target genes. A) Grey triangles represent distance from the TSS, while blue triangles depict the TSS of the individual genes. Direction of the arrow shows also whether the genes is transcribed on the + or – strand. B) Motif analysis identifies the binding motif for Zeb2 as CACCTG.
Figure S5. Genotyping of Zeb2ΔP/ΔP mESCs. A) Primers (listed in Table 1) used for genotyping. B) PCR confirming the proper deletion of the selected region in clone E9. C) Further PCR showing the difference in molecular weight for regions amplified with different set of primers and discriminating between WT, heterozygous (Zeb2ΔP/+) and homozygous (Zeb2ΔP/ΔP) deletion clones. D) Sanger sequencing showing the correct deletion of the selected region in Zeb2ΔP/ΔP clone.
Figure S6. Gene-to-Disease association of the D8 DEGs bound by Zeb2. The gene names of the DEGs expressed at D8 of neural differentiation were initially subjected to a mouse to human gene conversion. After that, gene to disease association analysis was performed showing that a number of those genes, could be associated with neurodevelopmental disorders, mental disorders, eye defects, seizures and speech impairment.
29 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Supplementary files
File S1. Narrow peaks obtained from the ChIP-seq.
File S2. Transcriptomic data of Zeb2 bound genes in differentiating mESCs.
File S3. Target genes bound by Zeb2 and deregulated in different already published RNA-seq datasets.
30 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Supplementary tables
Table S1: List of primers used in the study. Primer name Sense/Antisense Sequence (5’ -> 3’) Application FlV5mZeb2Ex9_Fwd Sense GGCTTACCTGCAGAGCATCA genotyping FlV5mZeb2Ex9_Rev Antisense CTCCATCTAACTCTGTCTTGGC genotyping FlV5_Fwd Sense CTACTCGCAGCACATGAATC genotyping FlV5_Rev Antisense GAGAGGGTTAGGGATAGGC genotyping ΔZP_P1_Fwd Sense GTCAGTCCGTCCCCAGGTTT genotyping ΔZP_P2_Rev Antisense GGCATGCTAGCTGGGCTGGT genotyping LN249_Fwd Sense GGAGCAAACTGAACAAAACCTCGCC genotyping LN249_Rev Antisense GGCGAGGTTTTGTTCAGTTTGCTCC genotyping LN209_Fwd Sense AGCGGATCAGATGGCAGTTCGCATG genotyping LN209_Rev Antisense CATGCGAACTGCCATCTGATCCGCT genotyping Zeb2_Fwd Sense CAATGCAGCACTTAGGTGTA qPCR Zeb2_Rev Antisense TTGCCTAGAAACCGTATTGT qPCR Zeb2V5_Fwd Sense GAAACGATACGGGATGAGGA qPCR Zeb2V5_Rev Antisense AGGAGAGGGTTAGGGATAGG qPCR Nanog_Fwd Sense TCT TCC TGG TCC CCA CAG TTT qPCR Nanog_Rev Antisense GCA AGA ATA GTT CTC GGG ATG AA qPCR Pou5f1_Fwd Sense AGA GGA TCA CCT TGG GGT ACA qPCR Pou5f1_Rev Antisense CGA AGC GAC AGA TGG TGG TC qPCR Sox2_Fwd Sense GCGGAGTGGAAACTTTTGTCC qPCR Sox2_Rev Antisense CGGGAAGCGTGTACTTATCCTT qPCR Pax6_Fwd Sense ACATCTTTTACCCAAGAGCA qPCR Pax6_Rev Antisense GGCAAACACATCTGGATAAT qPCR Acrv1b_Fwd Sense CTGCCTACAGACCAACTACACC qPCR Acrv1b_Rev Antisense CCACGCCATCCAGGTTAAAGA qPCR Lhx5_Fwd Sense AGAACCGAAGGTCCAAAGAA qPCR Lhx5_Rev Antisense TCACTTTGGTAGTCTCCGTA qPCR Ntng2_Fwd Sense CAAGGACTCTACGCTTTTCG qPCR Ntng2_Rev Antisense AGCACTCGCAGTCTTGAAAT qPCR Sema3f_Fwd Sense CTACACAGCATCCTCCAAGA qPCR Sema3f_Rev Antisense ACGGCATTCTTGTTTGCATT qPCR Smad1_Fwd Sense TACTATGAGCTCAACAACCG qPCR Smad1_Rev Antisense GAAGCGGTTCTTATTGTTGG qPCR Smad3_Fwd Sense CACGCAGAACGTGAACACC qPCR Smad3_Rev Antisense GGCAGTAGATAACGTGAGGGA qPCR
31 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Sox13_Fwd Sense CTTACAGGAGGTTGTGCCA qPCR Sox13_Rev Antisense TCCTTAGCTTCCACATTGCT qPCR Stat3_Fwd Sense CAATACCATTGACCTGCCGAT qPCR Stat3_Rev Antisense GAGCGACTCAAACTGCCCT qPCR Tcf4_Fwd Sense TTGAAGATGTTTTCGCCTCC qPCR Tcf4_Rev Antisense CCTGCTAGTCATGTGGTCAT qPCR Tgfbr2_Fwd Sense GAAGGAAAAGAAAAGGGCGG qPCR Tgfbr2_Rev Antisense TGCTGGTGGTGTATTCTTCC qPCR Amylase_Fwd Sense GGCTGAGTGTTCTGGGAT ChIP-qPCR Amylase_Rev Antisense CACGGTGCTCTGGTAGAT ChIP-qPCR Cdh1_R1_Fwd Sense GCTAGGCTAGGATTCGAACGAC ChIP-qPCR Cdh1_R1_Rev Antisense TGCAGGGCCCTCAACTT ChIP-qPCR
Table S2: shRNAs used. Name Sequence shZeb2_1 CCGGCCGAATGAGAAACAATATCAACTCGAGTTGATATTGTTTCTCATTCGGTTTTTG shZeb2_2 CCGGCCTCAGGAATTTGTGAAGGAACTCGAGTTCCTTCACAAATTCCTGAGGTTTTTG shZeb2_3 CCGGCCAGTGTCAGATTTGTAAGAACTCGAGTTCTTACAAATCTGACACTGGTTTTTG shZeb2_4 CCGGCCCATTTAGTGCCAAGCCTTTCTCGAGAAAGGCTTGGCACTAAATGGGTTTTTG shCTRL CCGGCAACAAGATGAAGAGCACCAACTCGAGTTGGTGCTCTTCATCTTGTTGTTTTT
32 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Supplementary Materials and Methods
ESC culture and differentiation
CGR8 (strain 129) wild-type and Zeb2-Flag-V5+ mESCs were cultured and differentiated towards the
neural lineage (Bibel et al., 2007, with few modifications). Briefly, mESCs were cultured on 0.1%
Gelatin-coated plates in ESC-medium: DMEM supplemented with 15% heat-inactivated (HI) FBS,
2mM L-Glutamine, 1x Non-Essential Amino Acids (NEAA), 143 µM β-Mercapto-Ethanol (β-EtSH)
(all ThermoFisher Scientific, TFS) and LIF (103 U/ml).
For ND, 4x106 cells were plated on non-adherent 10-cm dishes (Greiner) and allowed to form
cellular aggregates (CAs) in 10 ml CA-medium (DMEM, 10% HI-FBS, 2 mM L-Glutamine, 1x
NEAA, and 143 µM β-EtSH). From D4 of ND, cells were grown in CA-medium supplemented with 5
µM Retinoic Acid (RA). During the aggregation stages of ND, medium was changed every other day
by carefully collecting the aggregates with a 10-ml pipet and transferring them to a 15-ml conical tube.
The CAs were allowed to sink to the bottom of the tube where, after the previous medium was
carefully discarded, the CAs were then resuspended in fresh medium and transferred back to the
dishes. At D8 of ND, the aggregates were harvested and dissociated by resuspension in 1 ml Accutase
(TFS) and pipetting them up-and-down using a 1-ml pipet, after shaking them in a 37 C water bath
for 5 min. The Accutase was deactivated by adding 9 ml of fresh N2-medium (DMEM with 2 mM L-
Glutamine, 50 µg BSA/ml and 1x N2-supplement) to the dissociated cells and pelleting the cells
gently for 5 min at 200 g. The cells were resuspended in fresh N2. To ensure single-cell suspension,
the cells were filtered by passing them through a 40-mm nylon cell strainer (Corning). 2.5 x 105
cells/cm2 were plated on Poly-DL-Ornithine hydrobromide (Sigma) / Laminin (Sigma) coated plates.
Cells were harvested at D8 or D10 of ND.
Bibel M, Richter J, Lacroix E et al. Generation of a defined and uniform population of CNS progenitors and neurons from mouse embryonic stem cells. Nat Protoc 2007;2:1034-1043.
33 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Western blots
To check Zeb2-V5 protein, the 2BE3-clone ESCs were subjected to ND till D8. Cytoplasmic and
nuclear fractions were extracted using NePer-kit® (TFS). Protein concentrations were measured using
the Bradfort BCA (TFS) and equal quantities of protein lysates were loaded on 6% SDS-PAGs and
thereafter cut based on protein relative molecular mass. Gels were then transferred onto nitrocellulose
membranes (Amersham Bioscience), which were incubated overnight with anti-Zeb2 (Seuntjens et al.,
2009) and anti-V5 (Life Technologies) antibody, followed by incubation at RT with HorseRadish
Peroxidase (HRP) conjugated secondary anti-rabbit and anti-mouse antibodies, respectively (Jackson
ImmunoResearch). Protein bands corresponding to Zeb2 or Zeb2-V5 were visualized on an AI-600
digital imager (Amersham|). As loading control, we used Valosin-containing Protein (VCP) and anti-
VCP antibody (Santa Cruz sc-57492, mouse).
RNA extraction and RT-qPCR analysis
Total RNA was extracted from ESCs using TRI Reagent (Sigma), and used for cDNA synthesis with
RevertAid RT Kit (TFS) with oligodT-primers. RT-qPCR was performed using SybrGreen dye
(BioRad) on a CFX96 T1000 thermal cycler (BioRad). All data shown are averages of 3 independent
biological replicates and 3 technical replicates, normalized to β-Actin mRNA levels. Primers are listed
in Table 2. Analysis and data visualization was performed in R environment for statistical computing
version 3.5.3, implemented with the tidyverse version 1.3 package (https://github.com/tidyverse).
ChIP
For ChIP 2x108 cells were harvested at ND-D8 in 10 ml of PBS, and cross-linked at 1% formaldehyde
(Sigma Aldrich) for 15 min, rotating at RT. Quenching followed with 125 mM glycine during 5 min,
again rotating at RT. Cross-linked cells were washed twice with ice-cold PBS (5 min, 1,500 rpm (240
rcf), 4˚C), the pelleted cells snap-frozen, and stored at -80 C. For sonication, the cell pellets were
thawed on ice, resuspended in 1 ml sonication buffer (10 mM Tris-HCl pH 8.0, 1 mM EDTA, 0.5 mM
EGTA) supplemented with protease and phosphatase inhibitors (PPI, Roche) and incubated on ice for
10 min. DNA was sheared by sonicating the cells using a probe sonicator (32 cycles, 30 sec-on
34 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
amplitude 9, and 30 sec-off). These samples were centrifuged at 13,200 rpm (17,000 rcf) for 10 min at
4 C. Chromatin pellets were snap-frozen and stored at -80 C.
To check sonication efficiency, 50 µl of sample was de-crosslinked overnight by adding 5 mM
NaCl at 65 C, shaking (950 rpm, Eppendorf ThermoMixer C). The next morning 5, 10 and 20 µl of
sample were loaded on a 2% agarose gel, revealing ideally a DNA-smear around 300 bp. A 50-µl
sample was used as control input. For immunoprecipitation, chromatin of 107 cells was diluted in
ChIP-dilution buffer (17 mM Tris-HCl pH8, 170 mM NaCl, 1.2 mM EDTA, 0.01% SDS, 1.1% Triton
X-100, with 1xPPI) to a final volume of 1 ml. Samples were pre-cleared by adding pre-washed Protein
A/G agarose beads (Santa Cruz) and further incubation for 1 hour, rotating at 4 C. Then, samples
were centrifuged for 1 min (1,000 rpm; 106 rcf) at 4 C, and the pre-cleared chromatin (supernatant)
was transferred to a new low-binding 1.5-ml tube and incubated with 50 µl of pelleted V5-Agarose
beads (Sigma Aldrich), rotating at 4 C overnight. Before addition of V5-agarose beads, they were
washed 5 times (5 min each) in PBS by rotating them. As negative control, half of the sample was
incubated with Protein A/G beads (Santa Cruz, sc-2003).
The following day the beads were pelleted (1,000 rpm; 1 min) and washed as follows: once with
lower-salt buffer (20 mM Tris-HCl pH 8.0, 150 mM NaCl, 2 mM EDTA, 0.1% SDS, 1% Triton X-
100), transferred to non-stick low-binding 1.5-ml tubes and then washed once with high-salt buffer
(i.e. lower-salt buffer, but now 500 mM NaCl), once washed with LiCl buffer (10 mM Tris-HCl pH
8.0, 250 mM LiCl, 1 mM EDTA, 1% NP-40, 1% sodium deoxycholate (DOC), and twice washed with
10 mM Tris-HCl pH 8.0, 1 mM EDTA (each incubation for 5 min, rotating at 4 C, followed by
gently spinning down.
The protein-chromatin was then eluted from the beads by adding 250 µl of elution buffer (1%
SDS + 100 mM NaHCO3), rotating for 1 hour at RT twice, and combining the eluates from both steps.
To the input sample, 450 µl of elution buffer was also added, and all samples were de-crosslinked by
addition of 5 mM NaCl at 65 C overnight, shaking 950 rpm. The day-after, 2µl Proteinase-K (10
mg/ml), 20 mM (final concentration) Tris-HCl pH6.5, 5mM (final concentration) EDTA pH8.0 and 10
mg/ml RNase-A (Sigma) were added to each sample and incubated for 1 hour at 45°C while shaking
(700 rpm). DNA was extracted from the samples using the PCI method and diluted in water. Five
35 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
independent ChIPs were performed, for a total of 108 mESCs used per condition, and pulled-down
chromatin was pooled. ChIP efficiency was assessed by qPCR using primers amplifying Cdh1
promoter sequences bound by Zeb2 (Stryjewska et al., 2017). All primers used are listed in Table S1.
Zeb2 shRNA-mediated KD
Zeb2-KD in was done by transfecting Zeb2-shRNAs into mESCs at ND-D8. For this, the CAs were
dissociated as described above, and single-cell suspensions were transfected using Amaxa
Nucleofector II (using kit V, program A-33). In total 4 μg of shRNA was used for transfection of
4.5x106 cells. After transfection, ells were plated in 5 ml N2-medium on a poly-ornithine/laminin-
coated 6-cm cell culture dish. Two hours post-transfection, the medium was refreshed, and 24 hours
after transfection was changed to N2-medium+Puromycin (see above) for 48 hours. The cells were
then harvested, and KD efficiencies examined by RT-qPCR. As a control, scrambled shRNA was
used.
36 A Zeb2V5 ChIPseq etal2021_ Birkho C E 0.0 0.2 0.4 0.6
0.0 0.2 mRNA levels 0.4 0.6 bioRxiv preprint 0D 6D8 D6 D4 D0 0D 6D8 D6 D4 D0
CACCTG Zeb2 was notcertifiedbypeerreview)istheauthor/funder.Allrightsreserved.Noreuseallowedwithoutpermission. R1
CACCTG doi:
CACCTG https://doi.org/10.1101/2021.07.06.451350 ATG Zeb2-V5 WT transcripts Zeb2-Flag-V5 both alleles 0 1 2 3 4 5 D 0 1 2 3 4 F ; fold enrichment over input this versionpostedJuly6,2021. B (Agarose =1) D8 D6 D4 D0 115 140 0D 6D8 D6 D4 D0 10 20 25 15 80 0 5 Sox2 Nanog ECE NE
Zeb2V5 mESCs R1 pcDNA3-V5Zeb2 αVCP αV5 The copyrightholderforthispreprint(which 0.0 0.5 1.0 1.5 0 1 2 3 4 0D 6D8 D6 D4 D0 D8 D6 D4 D0 beads 115 140 80 αV5 Agarose Pou5f1 Pax6 ECE NE
Zeb2V5 mESCs
pcDNA3-V5Zeb2 αVCP αZeb2 Figure 1 Figure 2 A B
bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
protein coding genes (94%) 0.15
miRNAs (5%)
0.10 NA (1%) density
0.05
0.00 −10 −5 0 5 10 distance from TSS (kb)
C D
Wnt signaling pathway (P00057) Integrin signalling pathway (P00034) gene−specific transcriptional regulator (PC00264) Inflammation mediated by chemokine and cytokine signaling pathway (P00031) protein modifying enzyme (PC00260) Cadherin signaling pathway (P00012) transporter (PC00227) PDGF signaling pathway (P00047) EGF receptor signaling pathway (P00018) transmembrane signal receptor (PC00197)
Angiogenesis (P00005) protein−binding activity modulator (PC00095) VEGF signaling pathway (P00056) nucleic acid metabolism protein (PC00171) Transcription regulation by bZIP transcription factor (P00055) cytoskeletal protein (PC00085) Thyrotropin−releasing hormone receptor signaling pathway (P04394) TGF−beta signaling pathway (P00052) scaffold/adaptor protein (PC00226)
p53 pathway (P00059) membrane traffic protein (PC00150) Metabotropic glutamate receptor group III pathway (P00039) intercellular signal molecule (PC00207) Ionotropic glutamate receptor pathway (P00037) Interleukin signaling pathway (P00036) translational protein (PC00263) FGF signaling pathway (P00021) extracellular matrix protein (PC00102) Endothelin signaling pathway (P00019) chromatin/chromatin−binding, or −regulatory protein (PC00077) Cytoskeletal regulation by Rho GTPase (P00016) cell adhesion molecule (PC00069) Apoptosis signaling pathway (P00006) 0 5 10 0 2 4 6 %of genes % of genes
E Dmxl2 Wnt3a 32 Trim8 ... 61 58 Klf7 Gli2 Dnmt3l Myc Esrrb .... 335 Zeb2 Sema4c/6d/3f/5b .... 53 492
213 Tcf4 Camk2b Rara Cdk1 ....
Birkho et al 2020_ Zeb2V5 ChIPseq Figure 3 A Sema3f Fstl4 bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6,.... 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Nkx2.1|Zeb2 Gsh2|Zeb2 Available RNAseq Datasets 13 2294 peaks in Nkx2.1|Zeb2 39 38 (van den Berghe et al., 2013) protein coding regions = 1952genes 1 esiRNA KD of Zeb2 in ES cells (Cxcr4) (Dries et al., 2020) 1 2
Gsh2|Zeb2 (Deryckere et al., 2020) Zeb2 Smad3 14 Pax6
esiRNA KD of Zeb2 in ES cells
B
Signaling pathways regulating pluripotency of stem cells TGF−beta signaling pathway FoxO signaling pathway Hippo signaling pathway Axon guidance Adherens junction Cellular senescence AMPK signaling pathway Relaxin signaling pathway Apelin signaling pathway PI3K−Akt signaling pathway Endocytosis 0 5 10 15 20 −logP
Birkho et al 2020_ Zeb2V5 ChIPseq B mRNA levlels mRNA levlels mRNA levlels A 0.00 0.25 0.50 0.75 1.00 1.25 0.00 0.05 0.10 0.15 0.20 0.25 0.0 0.2 0.4 0.6 bioRxiv preprint
D0 D0 Smad2 D0 Cxcr4 Zeb2 ground-
D8 D8 D8 state 0D 6D D10 D8 D6 D4 D0 was notcertifiedbypeerreview)istheauthor/funder.Allrightsreserved.Noreuseallowedwithoutpermission. shCTRL D10 D10 D10 doi: https://doi.org/10.1101/2021.07.06.451350 D10 D10 D10 0.00 0.02 0.04 0.06 0.00 0.05 0.10 0.15 0.20 0.0 0.5 1.0 1.5 pLsEsNPCs EBs EpiLCs Smad3 D0 D0 D0 Nanog shZeb2 Lhx5
D8 D8 D8 Transfection D10 D10 D10
D10 D10 D10 Read-Out 0.00 0.01 0.02 0.03 0.04 0.0 0.1 0.2 0.3 0 1 2 3 4 ; this versionpostedJuly6,2021.
D0 Ntng2 D0 D0 Bmp7 Id2 D8 D8 D8 D10 D10 D10 D10 D10 D10 0.0 0.1 0.2 0.0 0.5 1.0 1.5 2.0 0.0 0.1 0.2 0.3 0.4 The copyrightholderforthispreprint(which Tgfbr2 D0 D0 Cited2 D0 Pax6 D8 D8 D8 D10 D10 D10 D10 D10 D10 0.00 0.03 0.06 0.09 0.12 0.00 0.25 0.50 0.75 0.0 0.2 0.4 0.6 Sema3f D0 D0 D0 Smad1 Tcf4
D8 D8 D8 Figure 4 D10 D10 D10 D10 D10 D10 A Figure 5
bioRxiv preprint doi: https://doi.org/10.1101/2021.07.06.451350; this version posted July 6, 2021. The copyright holder for this preprint (which 1 was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. log(FoldEnrichment)
76 589 4 3 2 1 Zeb2
44.99 mb 45.01 mb 45.03 mb 45.05 mb 45.07 mb 45.09 mb 45.11 mb
45 mb 45.02 mb 45.04 mb 45.06 mb 45.08 mb 45.1 mb
B
Zeb2 Nanog Bmp7 Tgfbr2 Smad1 0.80 0.80 0.80 6.00 0.60 0.60 0.10 0.60 0.40 4.00 0.40 0.40 0.05 2.00 0.20 mRNA levels 0.20 0.20 0.00 0.00 0.00 0.00 0.00 D0 D8 D10 D0 D8 D10 D0 D8 D10 D0 D8 D10 D0 D8 D10 Smad2 Smad3 Id2 Cited2 Sema3f 0.60 0.30 0.90 1.00 2.00 0.40 0.60 0.20 0.50 1.00 0.20 0.30 0.10 mRNA levels 0.00 0.00 0.00 0.00 0.00 D0 D8 D10 D0 D8 D10 D0 D8 D10 D0 D8 D10 D0 D8 D10 Cxcr4 Lhx5 Ntng2 Pax6 Tcf4 1.20 0.15 0.08 0.75 0.75 0.06 0.90 0.50 0.10 0.50 0.04 0.60 0.25 0.05 0.25 0.30
mRNA levels 0.02 0.00 0.00 0.00 0.00 0.00 D0 D8 D10 D0 D8 D10 D0 D8 D10 D0 D8 D10 D0 D8 D10
WT Zeb2ΔP/ΔP
Birkho et al 2021